mt5-small-finetuned-arxiv-cs-finetuned-arxiv-cs-full

This model is a fine-tuned version of shamikbose89/mt5-small-finetuned-arxiv-cs on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.4037
  • Rouge1: 39.8923
  • Rouge2: 20.9831
  • Rougel: 35.8642
  • Rougelsum: 35.8511

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5.6e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum
1.9675 1.0 500 1.5573 36.4989 18.4839 33.2984 33.2917
1.7523 2.0 1000 1.4972 37.7911 19.0357 33.5725 33.6058
1.6611 3.0 1500 1.4593 38.5822 19.4928 34.215 34.2531
1.6187 4.0 2000 1.4492 39.1219 20.8705 35.1969 35.2255
1.5864 5.0 2500 1.4289 39.7304 21.0654 35.6602 35.6667
1.5553 6.0 3000 1.4184 40.0696 21.0883 35.9536 35.9132
1.5215 7.0 3500 1.4163 39.1956 20.6757 35.5016 35.5196
1.5038 8.0 4000 1.4148 39.2373 20.3114 35.1676 35.1532
1.4929 9.0 4500 1.4064 39.9249 21.0155 35.8247 35.7937
1.4791 10.0 5000 1.4037 39.8923 20.9831 35.8642 35.8511

Framework versions

  • Transformers 4.12.5
  • Pytorch 1.10.0+cu111
  • Datasets 1.15.1
  • Tokenizers 0.10.3
Downloads last month
107
Hosted inference API
Summarization
Examples
Examples
This model can be loaded on the Inference API on-demand.